High-Affinity Olfactory Receptor for the Death-Associated Odor Cadaverine

Home , Cadaverine, Dimethylglycine, Isobutylamine, Spermidine

High-afﬁnity olfactory receptor for the death-associated odor cadaverine

Ashiq Hussaina,1,2, Luis R. Saraivaa,2,3, David M. Ferrerob,2, Gaurav Ahujaa,2, Venkatesh S. Krishnaa, Stephen D. Liberlesb,4, and Sigrun I. Korschinga,4

aInstitut für Genetik, Universität zu Köln, 50674 Cologne, Germany; and bDepartment of Cell Biology, Harvard University, Boston, MA 02115

Edited* by Cornelia I. Bargmann, The Rockefeller University, New York, NY, and approved October 18, 2013 (received for review October 2, 2013) Carrion smell is strongly repugnant to humans and triggers distinct Results innate behaviors in many other species. This smell is mainly carried Zebrafish Avoid a Cadaverine Odor Source. The zebrafish has in by two small aliphatic diamines, putrescine and cadaverine, which recent years emerged as an important model system for un- are generated by bacterial decarboxylation of the basic amino derstanding olfaction in vertebrates because of a remarkable acids ornithine and lysine. Depending on the species, these diamines similarity in the basic principles of olfactory representation (9) may also serve as feeding attractants, oviposition attractants, or and some technical advantages over the mammalian system (10). social cues. Behavioral responses to diamines have not been in- However, behavioral responses of zebrafish to diamines have not vestigated in zebrafish, a powerful model system for studying been described. We report here that zebrafish, like humans, show vertebrate olfaction. Furthermore, olfactory receptors that detect pronounced innate aversion behavior for cadaverine (Fig. 1). cadaverine and putrescine have not been identified in any species We developed a valence assay to measure behavioral re- so far. Here, we show robust olfactory-mediated avoidance behav- sponses of zebrafish to olfactory cues. Zebrafish were habituated to iorofzebrafish to cadaverine and related diamines, and concomi- a rectangular tank, and the position of each fish was recorded before tant activation of sparse olfactory sensory neurons by these diamines. and after odor delivery. Shifts in average position toward or away The large majority of neurons activated by low concentrations of from the odor source were recorded as attraction and avoidance, fi cadaverine expresses a particular olfactory receptor, trace amine- respectively. Food odor, an attractant for zebra sh, caused a mean P < associated receptor 13c (TAAR13c). Structure-activity analysis indi- displacement of 0.25 tank lengths (TLs, 0.01) toward the odor

B C NEUROSCIENCE cates TAAR13c to be a general diamine sensor, with pronounced source (Fig. 1 ), but tank water alone had no effect (Fig. 1 and Table S1). In contrast, cadaverine caused a mean displacement of selectivity for odd chains of medium length. This receptor can also P < be activated by decaying fish extracts, a physiologically relevant 0.28 TLs ( 0.01) away from the odor source, and was thus aversive (Fig. 1 A and C,andTable S1). Furthermore, the fish spent several- source of diamines. The identification of a sensitive zebrafish fold less time in close approach (distances < 0.05 TL) to the stimulus olfactory receptor for these diamines provides a molecular basis application site (P < 0.0001), although this area was not completely for studying neural circuits connecting sensation, perception, and avoided (P < 0.03) (Fig. 1A and Table S1), suggesting that the innate behavior. zebrafish did for short periods of time investigate the area, where stimulus was given. Mean velocity or total distance traveled was not Danio rerio | aversion | heterologous expression | polyamines altered during avoidance behavior (Table S1). Thus, the displacement observed is not caused by changes in motility but may result adaverine, putrescine, and other biogenic diamines are strongly from an assessment of odor valence by the fish. Crepulsive odors to humans, for whom these odors presumably Next, we analyzed whether related diamines were similarly signal bacterial contamination. It may be expected that animal aversive. We tested diamines with different carbon chain lengths, species feeding on carcasses attribute a more positive valence to ranging from C3 (diaminopropane) to C10 (diaminodecane). C diamines, and indeed both putrescine and cadaverine have been Avoidance behavior was observed (Fig. 1 ) to putrescine (C4), reported to be feeding attractants for rats (1) as well as goldfish (2). cadaverine (C5), diaminohexane (C6), diaminoheptane (C7), Similarly, insects depositing their eggs in carcasses or other pro- fi teineacous materials are attracted by these diamines (3). Beyond Signi cance signaling danger or food, putrescine and cadaverine also serve as social cues in several vertebrate species, both for marking of Cadaverine and putrescine, two diamines emanating from de- fl territories—for example, in feline species (4)—and for burial of caying esh, are strongly repulsive odors to humans but serve as conspecifics (5). innate attractive or social cues in other species. Here we show fi Very little is known about the molecular and cellular basis of that zebra sh, a vertebrate model system, exhibit powerful and cadaverine-driven behaviors. Cadaverine and putrescine evoke innate avoidance behavior to both diamines, and identify a high- electrophysiological responses in the olfactory epithelium of two affinity olfactory receptor for cadaverine. fish species (2, 6) and cadaverine-responsive olfactory sensory neurons and glomeruli have been identified in the mouse (7, 8). Author contributions: S.D.L. and S.I.K. designed research; A.H., L.R.S., D.M.F., G.A., and V.S.K. performed research; A.H., S.D.L., and S.I.K. analyzed data; and A.H., L.R.S., G.A., However, chemosensory receptors that detect cadaverine or re- S.D.L., and S.I.K. wrote the paper. lated diamines are unknown in any species, and could provide The authors declare no conflict of interest. valuable tools to study how the olfactory system mediates innate *This Direct Submission article had a prearranged editor. aversion or attraction. 1 Here, we show that cadaverine is a major product of zebrafish Present address: Department of Molecules, Signaling, and Development, Max-Planck- fi Institut für Neurobiologie, 82152 Martinsried, Germany. tissue decay, activates a zebra sh olfactory receptor (trace amine- 2 associated receptor 13c, TAAR13c) with high affinity, and elicits A.H., L.R.S., D.M.F., and G.A. contributed equally to this work. 3Present address: European Molecular Biology Laboratory–European Bioinformatics Insti- a strong, low-threshold, and olfactory-mediated avoidance response – fi fi tute (EMBL EBI) and Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, in zebra sh. In vivo measurements indicate that high af nity Hinxton-Cambridge CB10 1SD, United Kingdom. cadaverine responses occur primarily in TAAR13c-expressing 4To whom correspondence may be addressed. E-mail: [email protected] olfactory sensory neurons. These findings provide an important or [email protected]. foundation for understanding the molecular basis of a powerful This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. odor-driven behavior. 1073/pnas.1318596110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1318596110 PNAS Early Edition | 1of6 Downloaded by guest on September 26, 2021 Cadaverine A mean displacement 1.0 pre post (stimulus tracks)

Fig. 1. Aversive behavioral response of zebraﬁsh to x 0.5 diamines. (A and B) Cadaverine-evoked aversion and food odor-evoked attraction of individual zebraﬁsh

fish position (y) as visualized by movement patterns before (orange 0 tracks) and after (brown tracks) stimulus addition 0 0.25 0.50 0.75 1.0 (1 mM, 180 μL). The position of stimulus deposition fish position (x coordinate) B Food Odor (blue X), as well as the mean location before (blue 1.0 pre post (stimulus tracks) circle) and after (red, green circles) stimulus addition are indicated. The axes represent tank dimensions. (C) Mean displacement was expressed as a percent- age of tank length (n = 6 ± SEM). Addition of tank x 0.5 water alone had no signiﬁcant effect, whereas food odor elicited strong attraction. Similar avoidance

fish position (y) behavior was observed in male and female fish. 0 Significance was evaluated by Student t test, *P < 0 0.25 0.50 0.75 1.0 0.05, **P < 0.01. (D) Aversion behavior requires ol- C ** D ** E factory input. Nostril closure of zebrafish eliminates ** aversion to 1 mM cadaverine or putrescine. Black 30 ** 60 ** * ** bars, no treatment; white bars, nostril closure; **P < ** ** 20 40 45 ** 0.01, n = 3 ± SEM. (E) Avoidance behavior to dia- ** mines was dose-dependent. Responses to cadaverine 10 20

Aversion 30 and diaminoheptane were measured over a broad 0 0 ** concentration range (1 μMto1M,n = 3 ± SEM). The

* behavior is clearly saturable at 1-mM stimulus con- 10 Water 20 15 centration. Signiﬁcance was evaluated by Student Putrescine

anosmic anosmic < < Putrescine Cadaverine

Attraction t test, asterisks refer to cadaverine, *P 0.05, **P 20 Cadaverine 40

mean displacement (% of TL) mean displacement (% of Diaminoheptane mean displacement (% of TL) mean displacement (% of TL) mean displacement (% of 0.01. The slight decrease at higher concentrations is

Diaminooctane 0 Diaminohexane Diaminoheptane 0.001 0.1 10 1000 30 Diaminopropane not signiﬁcant except for putrescine (P < 0.05). (F)

Diaminodecane concentration (mM) Cadaverine-evoked aversion of an individual zebra- Food Putrescine ** Cadaverine fish in a flow-through two-channel set-up (11) as pre post (stimulus tracks) F 1.0 G * visualized by movement patterns before (orange ** tracks) and after (brown tracks) change of the upper channel to a chronic concentration of 10 μM cadav- 0.8 erine. The arrows indicate direction of flow (cadaverine, gray; water, white). Dotted gray lines enclose 0.4 the mixing zone not included in analysis of preference. The axes represent tank dimensions. (G)Aversion fi 0 behavior of zebra sh in the two-channel preference μ avoidance index avoidance test is maximal at 10 M cadaverine. The avoidance water cadaverine cadaverine water fish position (y coordinate) cadaverine index shows similar avoidance of the cadaverine -0.4 0 10 100 µM channel for 10 and 100 μM cadaverine (gray bars). 0 0 0.25 0.50 0.75 1.0 Significance in comparison with water was evaluated fish position (x coordinate) by Student t test, *P < 0.05, **P < 0.01.

and diaminooctane (C8), but not to diaminopropane (C3) or responses in this paradigm, with similar levels of avoidance ob- diaminodecane (C10). Time spent in close approach to the served at 10 μM (avoidance index 0.81 ± 0.18 SEM, n = 4, P < stimulus application site did not differ significantly from presti- 0.01) and 100 μM (avoidance index 0.73 ± 0.24 SEM, n = 4, P < mulus values for diaminopropane and diaminodecane, but was 0.05). Such sensitive detection suggests the existence of special- reduced to one-third or less for the other diamines (P < 0.0001) ized olfactory receptors for diamines. (Table S1). Only cadaverine and putrescine are naturally abundant in carrion (see below); other diamines not found ecologi- Cadaverine and Other Diamines Activate Sparse Olfactory Sensory cally may be aversive because they function as agonists for Neurons. We observed diamine-evoked increases in c-Fos expres- cadaverine and putrescine-activated receptors. sion and pERK levels in sparse olfactory sensory neurons. ERK Avoidance behavior to cadaverine and putrescine was abol- phosphorylation and c-Fos expression are induced by neuronal ished (P < 0.01) in fish, whose nostrils were closed by tissue glue, activity, and are widely used reporters for neuron responsiveness, showing the avoidance to be mediated by olfaction and not other including in the olfactory system (12, 13). Olfactory tissue was sensory modalities (Fig. 1D). Avoidance responses to cadaverine, obtained from zebrafish (n = 102) exposed to diamines or control putrescine, and diaminoheptane exhibited similar dose-dependence stimuli and stained using standard immunohistochemical (IHC) (Fig. 1E), with complete saturation at 1 mM diamine in the techniques. c-Fos expression was similarly and robustly induced stimulus, half-maximal values at 100 μM, and about one-third of by food odor, cadaverine, and other diamines (∼6.0 cells per maximal values at 10 μM(Fig.1E). The actual concentrations lamella) but not tank water alone (<0.5 cells per lamella), with encountered by the zebrafish at the time of decision-making are low background levels likely resulting from residual odors in tank expected to be much lower than these values because of dilution water (Fig. 2 A and B, and Fig. S1). The frequency of pERK- of a small stimulus volume (180 μL) into a large tank volume (9 containing cells was similar to that of c-Fos–expressing cells for L). To obtain a more stringent estimate of behavioral sensitivity, food odor, cadaverine, and diaminohexane. Some differences in we have therefore used a two-channel preference assay (11), with c-Fos induction and ERK phosphorylation were observed for a constant cadaverine concentration in one of the two streams other diamines, and could reflect differences in reporter sensi- during the stimulus period. Cadaverine evoked robust avoidance tivity for lower affinity ligands or the much longer exposure time

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1318596110 Hussain et al. Downloaded by guest on September 26, 2021 A amine receptors (21). However, most class I and class II teleost TAARs retain this amine-binding motif (17), making them good candidates for contributing to amine perception by the fish olfactory system. We initiated a chemical screen to identify agonists for zebra- Put Cad Hex Hep Oct fish TAARs and selected representatives from each of the five ** ** ** ** TAAR subfamilies retaining the amine-binding motif and three BC8 ** ** 6 D 6 ** TAARs without it (see Fig. 3 for phylogenetic position of genes 4 ** analyzed). We previously developed a reporter gene system to 4 ** 2 Cad 2 * Hep measure ligand-induced TAAR activation (16, 19); here, we used 0 this system to identify zebrafish TAAR agonists among 93 test 0 0.50 50 µM5 c-fos+ nuclei/lamella Water Food c-fos+ nuclei/lamella 193 876542 odorants, including a large number of amines, diamines, polyamines, D and amino acids. One receptor tested, zebrafish TAAR13c, gave robust responses to cadaverine (1,5-diaminopentane) and related aliphatic diamines (Fig. 3 A and C). Cadaverine activated HEK293 cells ex- Put Cad Hex Hep Oct pressing TAAR13c, but not control cells lacking TAAR13c, with a ** ** half-maximal response (EC )occurringat23± 3 μM (mean ± EF14 6 Put 50 12 Cad SD) and a threshold response occurring at 3 μM (Fig. 3E). Ca- ** 4 Hex 6 ** daverine variants, in which one amino group is replaced with ** 4 2 a hydroxyl group, methyl group, or hydrogen, did not activate ** ** ** B 2 ** TAAR13c (Fig. 3 ). Furthermore, 47 different aliphatic and ** ** 0 0 132 aromatic monoamines with varying chain lengths, degrees of Water Food pERK+ cells/lamella 193 876542 pERK+ cells/lamella 0 10 10 10 µM substitution, and functional groups, did not activate TAAR13c Fig. 2. Diamines elicit c-Fos and pERK increase in olfactory sensory neurons. (Fig. 3A). This structure–activity analysis suggests that TAAR13c (A) Zebrafish (n = 21) were exposed to stimuli indicated (2 or 5 mM). c-Fos contains two remote cation recognition sites, both of which re- IHC (green) and nuclear staining (propidium iodide, red), enabled visuali- quire occupancy for receptor activation. zation of c-Fos+ nuclei (yellow), some emphasized by gray arrow heads. (B and C) Quantification of c-Fos+ nuclei/lamella as a function of diamine chain

TAAR13c Preferentially Detects Odd-Chained Diamines. We next NEUROSCIENCE length (B) or concentration (C). Results from one representative experiment measured TAAR13c responses to diamines ranging from C3 each are shown. Counting was done on randomized micrographs, values to C10 across several concentrations (Fig. 3E) and found ca- given represent mean ± SEM. Significance in comparison with water was < < daverine (C5) and diaminoheptane (C7) to activate TAAR13c evaluated by Student t test, *P 0.05, **P 0.01. (B) 1, water; 2, food ex- fi = ± μ ± μ tract; 3–8, numbers reflect carbon chain length of diamines; 9, dia- with highest af nity (EC50 23 3 M and 30 2 M, re- fi = spectively). Diaminohexane (C6) had ∼fivefold reduced affinity, minodecane. (D) Zebra sh (n 14) were exposed to stimuli indicated > (1 mM). Some pERK-labeled cells (red) are emphasized by gray arrowheads; whereas putrescine (C4) and diaminooctane (C8) had 10-fold nuclear counterstain (DAPI, blue). Red central stripes in some panels, un- reduced affinity. Diaminopropane (C3) and diaminodecane (C10) specific label in the basal lamina outside the sensory region. (E and F) did not activate TAAR13c at any concentration tested. Further- + Quantification of pERK cells/lamella as function of chain length (E)or more, other dibasic ligands, including cystamine, agmatine, and concentration (F). Values given represent mean ± SEM. Significance in histamine, activated TAAR13c with reduced affinity (Fig. S1). comparison with water was evaluated by Student t test, **P < 0.01. (E) Although TAAR13c detected numerous primary amines, it Results from two experiments are shown; 1, water; 2, food extract; 3–8, showed reduced activity for the tertiary amine derived from numbers reflect carbon chain length of diamines; 9, diaminodecane. (F) putrescine, tetramethyl-1,4-diaminobutane (Fig. 3A). Indeed, Evaluation was partly on randomized data, no difference was seen between TAAR13c is phylogenetically closer to those mammalian TAARs randomized and nonrandomized evaluation. that detect primary amines than to those preferring tertiary amines (Fig. 3D)(cf.ref.18). Interestingly, both fish avoidance behavior and sensory neuron required for c-Fos expression. Dose-dependent analysis indi- responses showed no distinct preference for odd-chained dia- cated threshold c-Fos responses to cadaverine and diamino- μ C mines, suggesting the existence of other receptors tuned to even- heptane at 2 and 5 M, respectively (Fig. 2 ). Low concen- chained diamines. This finding is consistent with electrophysio- trations of cadaverine and putrescine resulted in very low F logical studies indicating limited cross-adaptation of olfactory frequencies of pERK-labeled cells (Fig. 2 ), consistent with responses to cadaverine and putrescine (2). Here, we identified detection by a single olfactory receptor (cf. refs. 14 and 15). The TAAR13c as a highly sensitive detector of odd-chained diamines range of diamine chain lengths (C3 to C10) that stimulate either that include the repulsive odor cadaverine. c-Fos expression or ERK phosphorylation in olfactory tissue includes all diamines that promote aversive behavior (C4 to C8), A Physiological Source of Cadaverine Activates TAAR13c. A physio- consistent with this behavior being mediated by olfaction. How- logically relevant source of diamine odors is decomposing car- ever, olfactory receptors that detect cadaverine in zebrafish or any casses, whose presence may signal danger. In these circumstances, species have not been identified so far. cadaverine will be present in a complex mixture together with other potentially odorous chemicals that also result from tissue TAAR13c Is an Olfactory Receptor for Cadaverine and Other Diamines. decay. Thus far we have examined TAAR13c responses to pure As in mammals (16), zebrafish TAARs function as olfactory re- chemicals; next we asked whether TAAR13c could detect cadav- ceptors (17). We reasoned that a zebrafish TAAR could mediate erine produced during the natural process of tissue decomposition. the cadaverine avoidance behavior because several rodent TAARs We prepared fish extracts by placing a killed zebrafish in PBS detect biogenic amines, including some highly aversive odors (16, for 0 min (“fresh”) or 1 wk (“rotten”). PBS solutions were ho- 18–20). As a result of numerous gene-duplication events, the mogenized and centrifuged to remove debris, and resulting zebrafish TAAR family is large, with 112 receptors encoded by supernatants used as “extracts.” We found that rotten fish extracts the zebrafish genome (17). So far, ligands have not been iden- provided a potent stimulus for TAAR13c, but that fresh extracts tified for any zebrafish TAAR, and identification of such ligands had no activity (Fig. 4A). Next, we analyzed the concentration would be a key step toward understanding their physiological roles. dependence of fresh and rotten fish extracts. We observed a half- The dynamic evolution of the teleost Taar gene family led to maximal response to rotten fish extracts diluted ∼1,000-fold from widespread loss (17) of an amine-binding motif found in biogenic the initial preparation (Fig. 4B). In contrast, fresh fish extracts

Hussain et al. PNAS Early Edition | 3of6 Downloaded by guest on September 26, 2021 A 12

4 (1 mM stimulus) Taar13c activity Taar13c

acid acidacid lysine valine aniline indole taurine uracil tyramine L- L-serine L- adenine gramine hexanalsucrose quinoline piperidine ß-alanine L-leucine histamine spermine creatinine riboﬂavin butylamine ethylamine quinaldine pyrrolidine L-arginine L-histidine tryptophan cystamine tryptamine octopamine hexylamine pentylamine L-isoleucine L-threonine spermidine ethanolamine benzylamine methylamine dibutylamine L-methionine L- ethyl butyrate isoamylamineisobutylamine 1-methylindole dimethylamine trimethylamine L-aspartic 5-aminoindole -naphthylamine isopropylamine L-glutamic L-phenylalanine 1,5-pentanediol 2-aminopentaneα cyclohexylaminecyclopentylamine agmatineethylenediamine sulfate -phenylethylamine1-methylpiperidine1-methylpyrolidine -aminobutyric acid 1,4-diaminobutane1,8-diaminooctane 1-amino-2-propanol3-methoxytyramine5-amino-1-pentanol2-methylbutylamine ß N,N-dimethylaniline 1,6-diaminohexane 4-aminobenzoic N,N-dimethylglycine1,10-diaminodecane1,5-diaminopentane1,7-diaminoheptane2,5-dimethylpyrazine5-methoxytryptamine

N,N-dimethylbenzylamine 4-methoxyphenethylamineN,N-dimethylethanolamine3-(methylthio)propylamine 1-(2-aminoethyl)pyrrolidine 1-dimethylamino-2-propanol N,N-dimethylisopropylamine 2-(dimethylamino)ethanethiol N,N-dimethylcyclohexylamineN,N-dimethylphenethylamine 4-(dimethylamino)butyric acid 3,4-dimethoxyphenethylamine Monoamines N,N-dimethyl-1-naphthylaminetetramethylammonium chloride Amino acids Diamines/Polyamines tetramethyl-1,4-diaminobutaneOthers 3-(dimethylamino)propiophenone

19 12 18 20 BC D 17 14 1 H N class III 6 2 NH2 16

2 HO NH2 8 15 class I 4 3 HC NH 3 2 23 21 M 10 4 HNHµ 2 4 2 25 11 Taar13c activity Taar13c (1 mM stimulus) Taar13c activity Taar13c 27 26 0 0 - 123 4 374865 1024 1 Cadaverine and related compounds Diamine carbon chain length Cm1 Cm2 3 E 1.0 2 Diaminoheptane H N NH 2 2 28 13 4 12 Cadaverine H2N NH2 NH2 Diaminohexane HN2 22 9 5 0.5 NH2 outgroup Putrescine H2N 6 7 NH2 8 class II Diaminooctane H2N TAAR13c activity TAAR13c Diaminopropane H N NH 2 2 Danio rerio taar NH2 genes 0 Diaminodecane H2N 0 1.0 10 100 1000 Branch support ≥ 99% concentration (µM) Amine-binding motif present

Fig. 3. TAAR13c is a sensitive diamine detector. (A) HEK293 cells were transfected with TAAR13c plasmid and a reporter gene, incubated with 93 different test chemicals (1 mM), and assayed for reporter gene activity. TAAR13c activity is reported after normalization to responses from control stimuli (media alone). Only aliphatic and mixed aliphatic/aromatic diamines and polyamines activate TAAR13c. No responses to diamines were observed in cells transfected with reporter gene alone. (B) TAAR13c activation requires a divalent ligand with two amino groups. TAAR13c responses (n = 3 ± SEM, 1-mM stimuli) were measured for: 1, cadaverine; 2, 5-amino-1-pentanol; 3, hexylamine; 4, pentylamine; or “–” no ligand. (C) TAAR13c prefers odd-chained diamines. Responses of TAAR13c (n = 3 ± SEM) were measured to diamines of carbon chain length 3–10 (100 μM). (D) Phylogenetic tree for taar genes, gene set as described previously (17), but only three mammalian species included (mouse, rat, human). Class I and II TAARs retain the amine-binding motif (yellow shade). Numbers indicate TAAR subfamilies, including mammalian TAARs that detect primary (blue) or tertiary (red) amines, as well as zebraﬁsh TAARs analyzed here (purple). TAAR13c terminal branch, light purple; Cm1, Cm2, elephant shark taar genes. (E) Dose-dependent activation of TAAR13c by aliphatic diamines; values are in relative units. A representative experiment is shown (n = 3 ± SEM). No responses to diamines were observed in cells transfected with reporter gene alone. Higher afﬁnity is seen for odd-chained diamines.

did not activate TAAR13c at any concentration tested, up to a 10- TAAR13 family members (Fig. S1). This antibody labeled a 55- fold dilution (Fig. 4B). These results indicate that TAAR13c is kDa protein in olfactory epithelium, but not other tissues by able to detect diamines in a complex and physiologically re- Western blot analysis, and an extremely sparse population of levant mixture. olfactory sensory neurons (0.3 cells per lamella) (Fig. S1) by IHC analysis. Two-color analysis indicated colabeling of olfactory Cadaverine Is the Principal TAAR13c Activator in Decayed Fish. To sensory neurons with TAAR13c antibody and Taar13c cRNA determine the most relevant TAAR13c ligands in rotten fish riboprobe, with Taar13c riboprobe labeling three- to sixfold extracts, we quantified the levels of diamines using liquid chro- matography and tandem mass spectrometry (LC/MS). The number more cells, likely because of cross-hybridization to the four other Taar13c of ion counts with the mass-charge ratio (m/z) corresponding to family members. Neurons labeled by TAAR13c anti- nine different diamines (putrescine, cadaverine, diaminohexane, body showed a ciliary morphology (Fig. S1) and occurred at a diaminoheptane, diaminooctane, agmatine, cystamine, histamine, frequency similar to that predicted for expression of individual and cysteamine) were separately graphed over time. The retention olfactory receptor genes (15). time and integrated area of observed peaks were compared with Next, we asked whether cells labeled by the TAAR13c anti- standards for chemical assignment and quantification. None of body responded to diamines using two-color IHC analysis for these nine amines were detected in fresh fish extracts, and only pERK and TAAR13c. We found that cadaverine-responsive cadaverine, putrescine, and histamine were detected in rotten fish neurons could be classified as low or high affinity based on re- extract (Fig. 4C). Cadaverine was the most abundant diamine sponse sensitivity. High-affinity (10 μM) cadaverine responses detected, and levels of cadaverine, but not putrescine or histamine, occurred primarily (∼90%) in TAAR13c-expressing cells (Fig. fi were suf cient to explain the striking sensitivity of TAAR13c for 5), whereas increasing cadaverine concentration 10-fold resulted decomposed tissue (Fig. 4D). in increased numbers of responsive neurons, suggesting recruit- fi ment of additional low-affinity receptors (Fig. 5). TAAR13c cells High-Af nity Cadaverine Responses Occur Primarily in TAAR13c- μ Expressing Neurons. We next sought to determine whether high- were distinct from those activated by low concentrations (10 M) + fi affinity cadaverine responses occurred primarily in TAAR13 of putrescine (Fig. 5), consistent with our ndings that TAAR13c − or TAAR13c olfactory sensory neurons. We generated a poly- prefers odd-chained diamines. Taken together, these results are clonal antibody that recognizes a highly divergent region of the consistent with TAAR13c being a major component in high-af- TAAR13c sequence that is not conserved in closely related finity cadaverine recognition by zebrafish.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1318596110 Hussain et al. Downloaded by guest on September 26, 2021 8 neurons show increased pERK levels at high cadaverine con- A B 0.8 rotten centrations, consistent with the presence of several olfactory receptors that can detect cadaverine. However, at low concen- 4 0.4 trations an extremely sparse population of receptor neurons (0.3

TAAR13c activity TAAR13c cells per lamella, about 100 cells per olfactory rosette) is acti- fresh 0 activity TAAR13c 0 fresh rotten 4 3 2 1 vated, corresponding to the lower limit of cell numbers found for 10 10 10 10 fold dilution individual olfactory receptor genes (cf. ref. 15). Moreover, the TAAR13c activators (mM) C in fish extract: fresh rotten D 1.0 large majority of these cells expresses TAAR13c, consistent with inferred Putrescine 4.0 cadaverine this receptor being a major component of high affinity cadav- rotten Cadaverine 16.1 erine detection. Further investigation, in particular TAAR13c Diaminohexane 0.5 loss-of-function analysis, will be required to delineate the exact Diaminoheptane Diaminooctane inferred role of TAAR13c in generating avoidance behavior to diamines. putrescine Cystamine The dose-dependence of cadaverine-evoked avoidance behav- TAAR13c activity TAAR13c Agmatine 0 4 3 2 1 iors, c-Fos and pERK induction, and TAAR13c activation are all Histamine 1.3 10 10 10 10 fold dilution understandably different, because stimulus application, signal de- Fig. 4. TAAR13c is activated by a biological source of diamines. (A) TAAR13c tection threshold, and signal/noise ratio are specifictoeach detects rotten but not fresh fish extracts diluted 100-fold. Values are given method. Receptor affinities can be much lower in heterologous as signal-to-blank ratios. (B) Rotten fish extract activates TAAR13c in a dose- systems than in vivo (28–30), where expression of receptors and dependent manner, whereas fresh extract shows no activity at any concen- signaling components are presumably optimized. Nevertheless, the tration. (C) LC/MS analysis showed that rotten but not fresh fish extracts threshold of cadaverine detection by TAAR13c of 3 μMisvery contain cadaverine, putrescine, and histamine at concentrations indicated. similar to in vivo thresholds observed for the intact olfactory sys- No other TAAR13c activators were detected (“–”, below detection limit of tem (6, 31) and to thresholds measured for isolated olfactory fi 0.5 mM). (D) TAAR13c activation by rotten sh extract can be explained by sensory neurons (2). Although it is difficult to estimate naturally cadaverine content. Gray solid curves indicate inferred TAAR13c activation occurring cadaverine concentrations close to dead fish, the ca- by cadaverine and putrescine present in the various dilutions of rotten fish fi extract and are superimposed on a curve (dashed lines) reporting measured daverine concentration we measured in rotten zebra sh extracts dose-dependent TAAR13c activation by rotten fish extract (see B). was several orders of magnitude higher than the TAAR13c activation threshold, and presumably high enough to allow detection of that odor source from some distance. Importantly, the Discussion behavioral response in the two-choice assay is maximal at the NEUROSCIENCE Cadaverine and putrescine are death-associated odors produced same low concentration of cadaverine, which elicits neuronal activity predominantly in TAAR13c-expressing cells. Thus, the by microbe-mediated decarboxylation of basic amino acids (22). fi Chemosensory receptors that detect these odors are unknown in available data are consistent with TAAR13c being a signi cant any species and could provide valuable tools to study how the ol- part of the receptor repertoire that detects cadaverine present in factory system mediates innate aversion or attraction (cf. ref. 23). ecologically relevant sources. Here, we show that zebrafish TAAR13c detects cadaverine TAAR13c arose during teleost evolution and orthologs are with high sensitivity and specificity. This study is unique in re- not found in rodents and humans who also detect cadaverine. porting a ligand for any of the 112 zebrafish TAARs, and fol- Thus, cadaverine-activated olfactory receptors in mammals may lowing identification of two amino acid-activated receptors from present a case of convergent evolution, either within the vertebrate TAAR family or between different olfactory receptor the V2R-related receptor family (24, 25), constitutes the third fi deorphanization of any fish olfactory receptor. TAAR13c is families (cf. ref. 27). In vitro studies did not identify a high-af nity strongly activated by primary amines and indeed is phylogenet- cadaverine receptor among mouse, rat, or human TAARs (18), ically closer to those rodent TAARs that prefer primary over although cadaverine reportedly activates TAAR-containing glo- tertiary amines (18). Moreover, TAAR13c has distinct molecular meruli in mice at high concentrations (8). Other mammalian recognition properties compared with many biogenic amine- TAARs also detect aversive amines; for example, isoamylamine activated G protein-coupled receptors in that it is selective for (TAAR3) and 2-phenylethylamine (TAAR4), both likewise diamines compared with monoamines. Structure activity analysis indicates an unusual divalent ligand binding pocket requiring two remote positive charges for activation. A conserved aspartic acid in biogenic amine receptors that forms a salt bridge with the li- 3.32 * gand amino group is retained in TAAR13c (Asp ), but resi- * * * dues important for recognition of the second amine are not * known. Nevertheless, the existence of a second amine contact site raises the possibility for a unique inverted mode of mono- amine recognition by G protein-coupled receptors that lose the TAAR13c pERK 10 µ M cad pERK 100 µ M cad pERK 10 µ M put conserved Asp3.32 but retain the second recognition site. TAAR13c is a narrowly tuned receptor that prefers medium- length, odd-chained diamines. Although a pronounced tuning to chain length is commonly found for olfactory receptors (26, 27), number of cells TAAR13c is peculiar in its strong preference for odd-chained cad put diamines. Odd- and even-chained diamines have significant dif- μM: 0 10 100 10 TAAR13c/pERK TAAR13c/pERK TAAR13c/pERK ferences in the relative orientation and positioning of the two amino fi groups, and their cognate olfactory receptors likely have negatively Fig. 5. Cadaverine activates TAAR13c-expressing neurons. Zebra sh were charged counterions in distinct locations of the agonist binding exposed to cadaverine (cad) and putrescine (put) at concentrations indicated and processed for concomitant IHC of TAAR13c (green fluorescence) and pocket. Interestingly, odd-chained and even-chained diamines did pERK (red fluorescence). DAPI was used as nuclear counterstain (blue). elicit comparable aversive behavior, which suggests the presence of Sometimes the basal lamina was stained unspecifically (green and red stripes fi + additional zebra sh olfactory receptors activated by even-chained in the center of some lamellae). Asterisks, pERK cells; arrows, colabeled cells + − diamines, consistent with data from cross-adaptation studies (6). (yellow) and pERK /TAAR13c cells (red). (Scale bars: 10 μm.) (Lower Left) Some estimates about the conceivable size of the cadaverine Quantitative evaluation, values are given as normalized cell numbers (120– receptor repertoire can be derived from our quantitative analy- 250 cells per condition were analyzed); green bar, TAAR13c+/pERK− cells; sis of sensory neuron responses. We find that many receptor yellow bar, double label; red bar, TAAR13c−/pERK+ cells.

Hussain et al. PNAS Early Edition | 5of6 Downloaded by guest on September 26, 2021 A pathway to aversion through cadaverine standard methods. For other clones, primer, and vector information, see SI Materials and Methods. TAAR13c Cadaverine TAAR Phylogenetic Analysis. The TAAR gene dataset was from ref. 17. For H 2 N NH2 algorithms used, see SI Materials and Methods.

TAAR Functional Assays. The reporter assay was performed as described previously (16). Details for assay and receptor sequences are provided in SI Aversion Materials and Methods.

• Zebrafish olfactory receptor TAAR13c is a cadaverine receptor Antibody Generation. A unique peptide of 16 amino acids served as immu- • TAAR13c prefers odd-numbered, medium length diamines • Decaying fish extracts activate TAAR13c due to cadaverine nogen for TAAR13c. Details are provided in SI Materials and Methods. • Cadaverine activates TAAR13c-expressing olfactory receptor neurons • Cadaverine triggers innate avoidance behavior in zebrafish. Western Blot. The Western blot was performed as described previously (34). Details are provided in SI Materials and Methods. Fig. 6. Graphical summary of key ﬁndings.

Immunohistochemistry. Standard procedures were used. Protocols are pro- produced by decarboxylation of amino acids (16, 18, 19, 32). Indeed, vided in SI Materials and Methods. amines are an odor group that is chemically suited both to aquatic Immunohistochemistry Combined with in Situ Hybridization. In situ hybrid- and airborne detection. Interestingly, trimethylamine, a TAAR5 izaiton-IHC was performed as described previously (35). Protocols and details agonist, is an aversive odor to humans and rats (20, 33) but attractive for antibody and probe are supplied in SI Materials and Methods. to mice (20). This finding is reminiscent of cadaverine, which is attractive to goldfish (2) but aversive to zebrafish (present results). Behavioral Analysis. Fish motion was tracked pre- and poststimulus addition. Taking these data together, we have shown that TAAR13c Two set-ups were used: single arena and two channel. Details of set-ups and emerges as a sensitive olfactory receptor for the death-associated analysis are provided in SI Materials and Methods. odor cadaverine, both in isolation and as part of a complex mixture. Cadaverine at low concentrations activates a sparse population of Quantitative LC/MS Analysis of Diamines. Protocols for sample preparation, TAAR13c-expressing olfactory sensory neurons and elicits power- processing, and analysis are provided in SI Materials and Methods. ful and innate avoidance behavior in zebrafish, a vertebrate model system. See Fig. 6 for a graphical summary of key findings. Such an ACKNOWLEDGMENTS. We thank Walter Nadler for help with programming; Yuichiro Oka for critical review of the manuscript; Wayne Korzan, Matthias association of odors and cognate receptors with a powerful avoid- Gruhn, and Kim Korsching for technical advice; Jamie Lemon and Priyanka Maity ance response provides a molecular basis for studying neural cir- for technical support; and Ansgar Bueschges for housing our behavioral set-up. cuits connecting sensation with perception of odor valence. This work was supported by Deutsche Forschungsgemeinschaft Award KO-1046/ 3 (to S.I.K.); an International Graduate School in Genetics and Functional Materials and Methods Genomics stipend (to A.H. and L.R.S.); an International Graduate School in Development Health and Disease stipend (to G.A.); a Boehringer Ingelheim TAAR Cloning. TAAR13c cDNA (National Center for Biotechnology Information travel grant (to L.R.S.); and a grant (Award Number R01DC010155) from the accession no. NM_001083040.1) was cloned from zebrafish genomic DNA using National Institute on Deafness and Other Communicative Disorders (to S.D.L.).

1. Heale VR, Petersen K, Vanderwolf CH (1996) Effect of colchicine-induced cell loss in 18. Ferrero DM, et al. (2012) Agonists for 13 trace amine-associated receptors provide the dentate gyrus and Ammon’s horn on the olfactory control of feeding in rats. Brain insight into the molecular basis of odor selectivity. ACS Chem Biol 7(7):1184–1189. Res 712(2):213–220. 19. Ferrero DM, et al. (2011) Detection and avoidance of a carnivore odor by prey. Proc 2. Rolen SH, Sorensen PW, Mattson D, Caprio J (2003) Polyamines as olfactory stimuli in Natl Acad Sci USA 108(27):11235–11240. fi – the gold sh Carassius auratus. J Exp Biol 206(Pt 10):1683 1696. 20. Li Q, et al. (2013) Synchronous evolution of an odor biosynthesis pathway and be- 3. Hamana K, Matsuzaki S (1984) Unusual polyamines in slime molds Physarum poly- havioral response. Curr Biol 23(1):11–20. – cephalum and Dictyostelium discoideum. J Biochem 95(4):1105 1110. 21. Huang ES (2003) Construction of a sequence motif characteristic of aminergic G 4. Burger BV, et al. (2008) Chemical characterization of territorial marking fluid of male protein-coupled receptors. Protein Sci 12(7):1360–1367. Bengal tiger, Panthera tigris. J Chem Ecol 34(5):659–671. 22. Mietz JL, Karmas E (1978) Polyamine and histamine content of rockfish, salmon, lobster, 5. Pinel JP, Gorzalka BB, Ladak F (1981) Cadaverine and putrescine initiate the burial of and shrimp as an indicator of decomposition. JAssocOffAnalChem61(1):139–145. dead conspecifics by rats. Physiol Behav 27(5):819–824. 23. Mori K, Sakano H (2011) How is the olfactory map formed and interpreted in the 6. Michel WC, Sanderson MJ, Olson JK, Lipschitz DL (2003) Evidence of a novel trans- – duction pathway mediating detection of polyamines by the zebrafish olfactory sys- mammalian brain? Annu Rev Neurosci 34:467 499. fi fi tem. J Exp Biol 206(Pt 10):1697–1706. 24. Speca DJ, et al. (1999) Functional identi cation of a gold sh odorant receptor. – 7. Nara K, Saraiva LR, Ye X, Buck LB (2011) A large-scale analysis of odor coding in the Neuron 23(3):487 498. olfactory epithelium. J Neurosci 31(25):9179–9191. 25. Demaria S, et al. (2013) Role of a ubiquitously expressed receptor in the vertebrate 8. Pacifico R, Dewan A, Cawley D, Guo C, Bozza T (2012) An olfactory subsystem that olfactory system. J Neurosci 33(38):15235–15247. mediates high-sensitivity detection of volatile amines. Cell Rep 2(1):76–88. 26. Fuss SH, Korsching SI (2001) Odorant feature detection: Activity mapping of structure 9. Yoshihara Y (2009) Molecular genetic dissection of the zebrafish olfactory system. response relationships in the zebrafish olfactory bulb. J Neurosci 21(21):8396–8407. Results Probl Cell Differ 47:97–120. 27. Saito H, Chi Q, Zhuang H, Matsunami H, Mainland JD (2009) Odor coding by a mam- 10. Lieschke GJ, Currie PD (2007) Animal models of human disease: Zebrafish swim into malian receptor repertoire. Sci Signal 2(60):ra9. view. Nat Rev Genet 8(5):353–367. 28. Mombaerts P (2004) Genes and ligands for odorant, vomeronasal and taste receptors. 11. Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V (2007) Smelling home can Nat Rev Neurosci 5(4):263–278. fi – prevent dispersal of reef sh larvae. Proc Natl Acad Sci USA 104(3):858 863. 29. Oka Y, et al. (2006) Odorant receptor map in the mouse olfactory bulb: In vivo sen- 12. Guthrie KM, Anderson AJ, Leon M, Gall C (1993) Odor-induced increases in c-fos sitivity and specificity of receptor-defined glomeruli. Neuron 52(5):857–869. “ ” mRNA expression reveal an anatomical unit for odor processing in olfactory bulb. 30. Krautwurst D, Yau KW, Reed RR (1998) Identification of ligands for olfactory re- Proc Natl Acad Sci USA 90(8):3329–3333. ceptors by functional expression of a receptor library. Cell 95(7):917–926. 13. Mirich JM, Illig KR, Brunjes PC (2004) Experience-dependent activation of extracel- 31. Friedrich RW, Korsching SI (1997) Combinatorial and chemotopic odorant coding in lular signal-related kinase (ERK) in the olfactory bulb. J Comp Neurol 479(2):234–241. the zebrafish olfactory bulb visualized by optical imaging. Neuron 18(5):737–752. 14. Weth F, Nadler W, Korsching S (1996) Nested expression domains for odorant re- 32. Kobayakawa K, et al. (2007) Innate versus learned odour processing in the mouse ceptors in zebrafish olfactory epithelium. Proc Natl Acad Sci USA 93(23):13321–13326. – 15. Sato Y, Miyasaka N, Yoshihara Y (2007) Hierarchical regulation of odorant receptor olfactory bulb. Nature 450(7169):503 508. fi gene choice and subsequent axonal projection of olfactory sensory neurons in ze- 33. Mitchell SC, Smith RL (2001) Trimethylaminuria: The sh malodor syndrome. Drug – brafish. J Neurosci 27(7):1606–1615. Metab Dispos 29(4 Pt 2):517 521. fi fi 16. Liberles SD, Buck LB (2006) A second class of chemosensory receptors in the olfactory 34. Ahuja G, et al. (2013) Zebra sh crypt neurons project to a single, identi ed medi- epithelium. Nature 442(7103):645–650. odorsal glomerulus. Sci Rep 3:2063. 17. Hussain A, Saraiva LR, Korsching SI (2009) Positive Darwinian selection and the birth 35. Oka Y, Saraiva LR, Korsching SI (2012) Crypt neurons express a single V1R-related ora of an olfactory receptor clade in teleosts. Proc Natl Acad Sci USA 106(11):4313–4318. gene. Chem Senses 37(3):219–227.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1318596110 Hussain et al. Downloaded by guest on September 26, 2021